Distortions and gender-related differences in the perception of mechanical engineering in high school students
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Pere Joan Ferrando Piera, Mar Gutiérrez-Colón Plana, Paloma Paleo Cageao, Silvia de la Flor López and Francesc Ferrando Piera ISSN 0214 - 9915 CODEN PSOTEG Psicothema 2013, Vol. 25, No. 4, 494-499 Copyright © 2013 Psicothema doi: 10.7334/psicothema2013.15 www.psicothema.com Distortions and gender-related differences in the perception of mechanical engineering in high school students Pere Joan Ferrando Piera, Mar Gutiérrez-Colón Plana, Paloma Paleo Cageao, Silvia de la Flor López and Francesc Ferrando Piera Universidad Rovira i Virgili Abstract Resumen Background: The reason for this study was the low interest that high Distorsiones y diferencias de género en la percepción de la ingeniería school students, particularly females, show for the subject of mechanical mecánica en alumnos de Bachillerato. Antecedentes: la motivación del engineering (ME). We assumed that this problem was partly due to: (a) presente estudio es la baja demanda de la carrera de ingeniería mecánica lack of understanding of the tasks involved in ME, and (b) a distorted por parte de los alumnos, y muy especialmente las alumnas, que ingresan and negative perception of the professional environment and working en la Universidad. Se planteó que, en parte, esta situación se debía a: (a) conditions. Method: To assess these two assumptions, two measurement un conocimiento erróneo de las tareas propias de la profesión, y (b) una instruments (tasks and perceptions) were developed and administered percepción distorsionada y negativa de su entorno y condiciones de trabajo. in a sample of 496 high school students. A multiple-group design was Método: para evaluar estos dos puntos se desarrollaron dos instrumentos used and data was analyzed by using an extended item response theory de medida (tareas y percepciones) que se administraron en una muestra de model. Results: In general terms, the results agreed with our expectations. 496 alumnos de Bachillerato. Se utilizó un diseño multi-grupo y los datos However, no significant gender differences were found. Discussion: The se analizaron de acuerdo a un modelo extendido de la teoría de respuesta al implications of the results for future improvements are discussed. ítem. Resultados: en términos generales los resultados apoyan los supuestos Keywords: Mechanical engineering and gender differences, attitude de partida. Sin embargo, no se encontraron diferencias significativas entre measurement, multi-group factor analysis, item response theory. hombres y mujeres. Discusión: finalmente, se discuten las implicaciones de los resultados cara a futuras acciones de mejora. Palabras clave: ingeniería mecánica y diferencias de género, medición de actitudes, análisis factorial multi-grupo, teoría de respuesta a los ítems. In recent decades, a considerable decrease has been observed stabilized around values below 10% (see Ferrando, 2012, for a in the demand for STEM (science, technology, engineering and summary). mathematics) university degrees in developed countries in general The decrease in the demand of the degree together with the lack (e.g., Byars-Winston & Canetto, 2011; Knight et al., 2012; NSRCG, of interest shown by women is creating a serious problem. For 2013) and Spain in particular (CRUE, 2008; Engynicat, 2008). example, it is not clear that the existing engineering workforce in the Furthermore, this trend seems to be specially pronounced in the USA will be able to maintain the present level of competitiveness case of engineering degrees, particularly mechanical engineering (NSRCG, 2013). The problem has been studied by a wide range of (NSRCG, 2013; Ferrando et al., 2012). research projects, most of which have two clear purposes: (a) to Apart from this general decrease, mechanical engineering assess the situation and identify the main sources of the problem, (ME) suffers from a well-known specific problem: the lack of and (b) to propose improvements. At the international level, these interest shown by female students (Byars-Winston & Canetto, projects have been initiated either by the government or private 2011; Hill, Corbett, & Rose, 2010; Medina, 2004; Knight et al., foundations (for example, the ADVANCE initiative). In Spain, 2012). In Spain, the importance of this specific problem is quite most of them have been carried out by universities: for example, clear (e.g., CRUE, 2008). After a timid increase in the 1970s, the the “Enginycat” (2008) project in Catalonia, the “Valentina” percentage of women studying ME in Spanish universities has program in Valencia (Díez, 2010), and the proposals by Medina (2004) in Jaen and del Río (2009) in Madrid. The reviewed proposals, however, do not focus specifically on ME but on general Received: January 11, 2013 • Accepted: May 17, 2013 engineering or, even more broadly, on STEM degrees in general. Corresponding author: Pere Joan Ferrando Piera The present study is also part of a research project that has been Facultad de Ciencias de la Educación y Psicología carried out at the Rovira i Virgili University (URV) and which Universidad Rovira i Virgili 43007 Tarragona (Spain) focuses specifically on ME. The purposes of the project are: (a) to e-mail: perejoan.ferrando@urv.cat make a diagnosis of the situation, and (b) use the results to make 494
Distortions and gender-related differences in the perception of mechanical engineering in high school students a series of proposals for high schools designed to increase the To close this section, we shall summarize the main contributions demand of the degree, particularly among women. In this paper we of the present study. First, our study focuses specifically on ME. shall discuss the first purpose, and the diagnosis will be made using Second, most previous studies focused on external sources: for a psychometric approach. example, stereotypes, gender-related prejudices, differential work The initial predictions of the study were made on the basis of (a) and family demands (Del Río, 2009; Hill, Corbett, & Rose, 2010; the experience of the ME teachers at the URV and (b) the results of Medina, 2010; Spencer, Steele, & Quinn, 1999). In contrast, our a series of semi-structured interviews administered to women who starting points focus on knowledge and perception of ME as a were studying the ME degree or who were professional engineers career. So, we believe that the present results can complement (Ferrando et al., 2012). The general starting point was that the low those obtained in previous studies. Finally, our study uses a demand was partly because of an incomplete or distorted perception rigorous psychometric methodology and provides measurement of ME as a career. This point can be divided into two more specific instruments that can be used in future studies in Spain. assumptions. First, high school students’ knowledge about the competences and tasks of ME is at least incomplete and perhaps Method largely incorrect. Second, these students’ perception of the work environments of ME professionals is negatively distorted, and this Participants and procedure distortion is expected to be more pronounced in women. As far as the first assumption is concerned, we note that in Spain, the image The sample was collected from ten high schools in the province of ME is closely related to the world of cars and motorcycles, of Tarragona, and can be considered to be representative of the particularly competition. In fact, however, only a small percentage students that are admitted to the URV every year. The initial sample of mechanical engineers work in this environment. And with consisted of 496 high school students aged between 16 and 18, regards to the second point, we believe that the work environments who were studying science or technology subjects. As explained of ME are perceived as dirty, dangerous, physically demanding, below, inconsistent response patterns were removed from the data, and possibly hostile (see also Medina, 2004). Again, however, so that the final sample consisted of 408 participants: 159 women most mechanical engineers work in offices. and 249 men. It is noted that the percentage of respondents of each The psychometric assessment of the two above-mentioned gender in the trimmed sample was similar to that in the original specific points prompted us to develop two scales that we shall sample (about 39% women). call “Tasks” (T) and “Perceptions” (P). Items on the T scale refer Questionnaires were administered in paper and pencil format in to tasks or activities that may or may not be specific to ME, and classroom groups and always by the same examiner. Administration the respondent must decide whether or not they are. Items on the P was voluntary and anonymous: the only data requested were scale are more general, and refer to the work environment and the gender and age. personal characteristics of ME professionals. As well as the degree of specificity, the scales also differ in their degree of concreteness- Instruments abstraction (the P items are more abstract) and in the denotative- connotative dimension (Osgood, Suci, & Tannembaum, 1957), Items on the T (18 items) and P (9 items) scales were developed because P items reflect respondents’ mental representation of ME by one of the authors on the basis of (a) his experience as head as a career more than T items. The main difference, however, of the ME degree, (b) the results of preliminary assessments to is that the T scale is mainly a maximum-performance objective students, and (c) the semi-structured interviews referred to above. measure (Cronbach, 1990) because its responses are naturally The content validity of the T items can partly be objectively scored as correct-incorrect (competences in ME are defined by assessed because, as mentioned above, the competences of a law). In contrast, the P scale is more similar to a typical-response mechanical engineer are defined by law. The content validity of measure (Cronbach, 1990) and can be considered to be an attitude the T and P items was also assessed using inter-judge agreement. A scale. This last distinction, however, is not so clear-cut because the committee made up of three experts (all of whom were mechanical perceptions endorsed in the P scale can be realistic or distorted to engineers) assessed the initial pool of items, and the chosen items some extent. were those in which complete agreement was reached. The content Two basic requirements are needed if T and P measures are to of these items can be seen in Table 2. be used to assess (a) knowledge and perception at the individual In both scales, the response format is YES-NO. One item in each and group level, and (b) potential gender-related differences in scale (T1 and P7) was used solely to assess response consistency these two variables. First, they must both behave as essentially and was not included in further analysis. On the basis of points (a) unidimensional measures and have some degree of measurement and (b) above, the content of these ‘validity’ items was considered accuracy. Second, ideally their items should be gender invariant. too obvious to a high school student studying scientific or technical Differential item functioning (DIF, see Muñiz, 1990) would mean subjects. So, erroneous or distorted responses to the ‘validity’ items that these items have different measurement properties in men were considered to be indicators of response inconsistency, and the and women, which would make gender-related comparisons more corresponding pattern was omitted from the data matrix. complex. Overall, then, the study had to be carried out in four stages: (a) development of the T and P measures, (b) assessment Data analysis of their psychometric properties (unidimensionality, accuracy and invariance) in a representative sample, (c) comparison of the Given the different characteristics of the T and P scales, they were estimated levels in knowledge and perception in men and women, analyzed separately, and the psychometric model for the analysis and (d) assessment of responses to individual items in order to was determined on the basis of the response format (binary) and identify the main sources of distortion and lack of knowledge. the purposes of the study. In both T and P, the chosen model was 495
Pere Joan Ferrando Piera, Mar Gutiérrez-Colón Plana, Paloma Paleo Cageao, Silvia de la Flor López and Francesc Ferrando Piera the multiple-group extension of the two-parameter normal-ogive Table 1 model (Muthén & Christofferson, 1981). Even though this model is Multiple-group analysis: Goodness of fit of the strong invariance model and most commonly used in its item response theory parameterization group structural parameter estimates (means) formulation, we decided to use the factor-analytic formulation for (a) T Scale three reasons. First, the model-data fit assessment of the factor model the unidimensionality assumption to be rigorously assessed. Model χ2 df RMSEA 90% C.I. CFI NNFI Second, the invariance in the item measurement properties (i.e., Strong I 492.51 254 0.065 (0.056;0.074) 0.92 0.92 lack of DIF) is also better assessed by using this modeling. Finally, the model provides estimates of the structural parameters at the Women Men group level (i.e., group means and variances on the corresponding θ mean and 90% C.I. 0 (fixed) 0.011 (-0.219;0.241) traits) that enable gender differences to be strictly assessed. In the factor-analytic parameterization of the multiple-group (b) P Scale model, between-group differences in thresholds/intercepts are 2 considered to indicate uniform DIF, whereas differences in the Model χ df RMSEA 90% C.I. CFI NNFI loadings indicate non-uniform DIF (e.g., Hernández & González- Strong I 60.82 46 0.040 (0.000;0.065) 0.97 0.97 Romá, 2003). In accordance with this criterion, we consider that between-group invariance in both thresholds and loadings is a Women Men sufficient condition for assuming that the measure under scrutiny θ mean and 90% C.I. 0 (fixed) 0.159 (-0.121;0.439) is free from DIF (see Little, 1997). This condition is known as “strong invariance” in the factor-analytic literature (Millsap & Meredith, 2007). In the binary case, fitting a multiple-group model with the strong in this case they provide more information than the corresponding invariance restrictions allows the relative means in each group to thresholds. Finally, given that the condition of strong invariance be estimated (in fact the model is over-determined for this purpose) is met and that there are no significant group differences, only a and, therefore, the potential gender differences to be assessed. The single pooled estimate of each parameter common to both groups most usual procedure consists of setting one of the means to zero is reported. (in our case that of women’s group) and freely estimating the mean Before the factorial solutions are interpreted, it is noted that of the other group together with its corresponding standard error overall measurement accuracy is acceptable in both scales even (Muthén & Christofferson, 1981). though some loadings (particularly in T) are quite low. The The model described above was fitted separately to T and P data reliability estimates of the factor scores were 0.83 (T) and 0.80 by using WLSM estimation as implemented in the Mplus version (P). 5.1 Program (Muthén & Muthén, 2007). As well as the chi-squared The T structure can only be interpreted if it is taken into account statistic, the following goodness-of-fit indices were used: (a) root that all of the items have been scored in the same direction: higher mean square error of approximation (RMSEA) with its 90% scores mean a higher level of knowledge. So, the difficulty indices confidence interval (Browne & Cudeck, 1993), (b) comparative are the proportions of correct responses in the corresponding items. fit index (CFI), and (c) nonnormed fit index- Tucker-Lewis index To start with, it seems clear that the structure is “positive manifold”, (NNFI-TLI) (see Hu & Bentler, 1999). as is to be expected if all items measure a general dimension of knowledge in this domain. Closer inspection, however, reveals Results interesting results. First, there is a cluster of items with both high loadings and high proportions of correct responses (3, 5, 13, 15, Table 1 shows the model-data fit results for both measures as 16 and 17) in which the level of knowledge seems to be clear. well as the mean estimates in each group. According to the literature The content of these items refers to manual tasks specific to the (Browne & Cudeck, 1993; Hu & Bentler, 1999; MacCallum & profession of mechanic and even blacksmith, which respondents Austin, 2000), the model-data fit can be regarded as excellent in consider to be related to ME (high loadings) but, at the same time, the case of P and acceptable in the case of T. Overall, we consider understand that they are not actually tasks of an engineer. that the strong invariance condition is met in both scales. So, we At the other extreme there is a cluster of items with very low conclude that there are no items with gender-related differential loadings and high proportions of incorrect responses. In principle, functioning or, more conceptually, that men and women interpret low discriminations might indicate problems of understanding due the items in the same way, and that, as measures, the items behave to item complexity, ambiguity, abstractness or length (Ferrando essentially in the same way in both groups. & Demestre, 2008). Inspection of the stems, however, suggests The structural results do not support our starting assumptions that this is not the case. Rather, it seems more plausible to simply on gender-related differences. Confidence intervals around the assume that respondents perceive these items as being far removed means which are freely estimated clearly show that there are no from the dimension that is measured. The most extreme stems significant differences in any case. So, the results suggest that both in this cluster refer to: organizing special transports (Item 8), the knowledge of tasks and the perceptions of work environments organizing the manufacturing process of computer keywords (Item are essentially the same in men and women. 9) and, above all, designing the trial for an artificial knee (Item 11). For both T and P, Table 2 shows the item measurement parameter In these items, the proportions of incorrect responses are around estimates: discriminations and difficulties. Discrimination 60-70%. So, overall, these items are not only probably perceived parameters are the standardized weights or loadings (λ). As for as quite unrelated to a career in ME but many respondents also difficulties, we decided to report the means or proportions because believe that the corresponding tasks do not belong to ME. 496
Distortions and gender-related differences in the perception of mechanical engineering in high school students Table 2 Item parameter estimates based on the strong-invariance model T Scale Item Content λzc Pc T1 Diseñar un motor* [ Designing an engine] – – T2 Proyectar una lavadora [Designing a washing machine] 0.20 0.62 T3 Cambiar el aceite de un automóvil [Changing the oil of a car] 0.79 0.80 T4 Apretar tornillos de máquinas [Tightening screws of a machine] 0.75 0.64 T5 Soldar tuberías [Welding pipes] 0.64 0.80 T6 Calcular la estructura de una nave industrial [Calculating the structure of an industrial warehouse] 0.30 0.74 T7 Proyectar ascensores [Designing elevators] 0.15 0.86 T8 Organizar y dirigir transportes especiales [Organizing and directing special transport] 0.07 0.35 T9 Organizar la fabricación de teclados de ordenador [Organizing the manufacture of computer keywords] 0.13 0.39 T10 Proyectar instalaciones de aire acondicionado [Designing air-conditioning facilities] 0.10 0.60 T11 Definir el ensayo de una prótesis de rodilla [Defining the trial of an artificial knee] 0.02 0.27 T12 Revisar las máquinas de un gimnasio [Revising gym equipment] 0.63 0.80 T13 Cambiar personalmente la pala de una excavadora [Personally changing a bulldozer blade] 0.70 0.78 T14 Pulir elementos metálicos [Polishing metal components] 0.68 0.72 T15 Desatascar tuberías de una vivienda [Unblocking pipes at home] 0.73 0.93 T16 Golpear con un martillo [Using a hammer] 0.84 0.90 T17 Engrasar engranajes [Greasing gears] 0.84 0.70 T18 Arreglar el reloj de un campanario [Mending the clock in a clock tower] 0.38 0.68 P Scale Item Content λzc Pc Hace falta tener un carácter especialmente fuerte para dirigir equipos de obreros poco cualificados [You need to have a specially strong P1 -0.14 0.59 character to direct teams of unskilled workers] P2 Se trabaja casi siempre en entornos con riesgo de accidentes físicos [You almost always have to work in physically dangerous environments] -0.64 0.67 P3 Se puede uno dedicar al diseño de productos [You can dedicate your time to designing products] 0.49 0.15 P4 Se trabaja casi siempre en entornos sucios [You almost always have to work in dirty environments] -0.73 0.83 P5 Casi siempre hay que llevar como vestido un mono azul [You almost always have to wear blue overalls] -0.84 0.83 P6 Se trabaja mucho con ordenadores[ [You do a lot of work on the computer] 0.50 0.14 P7 Se utilizan modelos físico-matemáticos para desarrollar el trabajo * [You have to use physical and mathematical models to do the work*] – – Hace falta una fortaleza física importante para desarrollar las funciones propias del oficio [You need to be very strong physically to work as a P8 -0.58 0.80 professional mechanical engineer] P9 El entorno de trabajo principal son despachos y oficinas [You work largely in offices] 0.64 0.52 Note. λzc: pooled standardized loadings; Pc: pooled proportion of endorsement ; *: validity item The P solution also agrees with the expectations and, Discussion furthermore, the distribution of the algebraic signs of the loadings clearly suggests a bipolar structure. In accordance with This study provides results that, as intended, may give rise to our initial views, we interpret a positive pole related to ‘clean’ actions of improvement. Before discussing them, however, we shall environments (computer-assisted design, office work) and a focus on some of the limitations to the generalizability of the study. negative pole related to potentially hostile, dangerous and ‘dirty’ First, the sample used here can be considered to be representative of environments. Furthermore, inspection of item discriminations the population of high school students in the province of Tarragona and difficulties (proportions of item agreement) suggests studying science and technology subjects. However, there is no that, in this group of respondents, the negative perception is guarantee that this sample is also representative of the entire considerably stronger than the positive perception. In particular, Spanish population. Generalization of the results is a matter for note that the strongest loading is on the idea of ‘blue overalls’. further research and, in this regard, the measurement instrument Also remarkable is the percentage of respondents who relate developed in this study can be a useful tool. ME to physically dangerous (70%) and “dirty” (80-85%) The fact that the sample used was restricted to the students environments. studying science and technology is also a limitation, because 497
Pere Joan Ferrando Piera, Mar Gutiérrez-Colón Plana, Paloma Paleo Cageao, Silvia de la Flor López and Francesc Ferrando Piera much richer results might have been obtained by considering and there appear to be no significant gender-related differences on comparing all the high school specialties. In this regard, we note these points. So, in principle, our results cannot explain (at least that the study arose out of the need to improve a situation. So, we directly) the lack of interest women show for ME. One possible focused only on the potential ‘consumers’ of our subject. explanation, which focuses on the P results, is that the ‘negative’ The results obtained here can be summarized in three main view differentially affects men and women. In other words, points. First, there is a cluster of tasks traditionally related to ME both men and women tend to perceive ME as ‘dirty’, ‘hard’, that respondents tend to identify quite clearly. However, in other ‘dangerous’ and ‘blue-collar’. However, this perception does not not so obvious tasks, the general lack of knowledge is considerable. particularly worry men, but women find it sufficiently important Second, the perception of work environments and characteristics not to seriously consider ME as a career. Indeed, this warrants related to a career in ME is generally negative, and appears to be further research. associated with stereotypes that are no longer real. Finally, contrary To sum up, we believe that the results discussed so far are to expectations, no noticeable gender-related differences appear to directly linked to potential actions of improvement to be carried exist regarding tasks and perceptions. out in the second stage of the project. As far as the tasks of a The first two points above agree with the expectations of the mechanical engineer are concerned, if information were provided study. However, the results obtained add new and useful information. about computer-assisted design, organization, transport, and Thus, although the knowledge of ‘traditional’ tasks is generally biomechanics, the interest shown by students in general and acceptable, we note, for example, that about 40% of respondents women in particular might increase. After all, these tasks are believe that designing a washing machine is not a task for a neither ‘dirty’ nor ‘blue-collar’. And as far as the perceptions of mechanical engineer. However, it clearly is: here, we are referring mechanical engineering are concerned, action should be taken to to an industrial machine. As for the “not so obvious” tasks, lack of break away from the negative stereotypes and to spread the real knowledge is mostly associated with the most usual tasks of ME image of a modern-day career in mechanical engineering. at present (transport and manufacturing) or the ones that are most active or that have the brightest future (biomechanics). As regards Acknowledgements the second point, the view of a career in ME is negatively distorted, as expected. However, it is interesting to note that the negative views This study has been partially supported by a fund from the mostly focus on the ideas of danger, physical demands, “dirtiness” “DASC - Departament d’Acció Social i Ciudadanía, Institut Català and, above all, the belief that ME is a blue-collar profession. de les Dones de la Generalitat de Catalunya” within the program Even though respondents tend to show a considerable lack of “Subvencions per a treballs de recerca en matèria d’estudis de knowledge and a negatively distorted view of the career in general, gènere i de les dones.” Ref U-52/10. References ADVANCE: http://www.mtu.edu/advance/. differential item functioning in polytomous ordered items. Psicothema, Browne, M.W., & Cudeck, R. (1993). Alternative ways of assessing model 15, 322-327. fit. In K.A. Bollen & J.S. Long (Eds.), Testing structural equation Hill, C., Corbett, C., & Rose, A., (2010). Why so Few? Women in Science, models (pp. 445-455). Newbury Park, CA: Sage. Technology, Engineering and Mathematics. AAUW: Washington. Byars-Winston, A., & Canetto, S. (2011). Preface: Accomplishments and http://www.eric.ed.gov/PDFS/ED509653.pdf. challenges for a diversity of women in science, technology, engineering Hu, L.T., & Bentler, P. M. (1999). Cut-off criteria for fit indexes in covariance and mathematics education and occupations. Journal of Women and structure analysis: Conventional criteria versus new alternatives. Minorities in Science and Engineering, 17, 1-3. Structural Equation Modeling: A Multidisciplinary Journal, 6, 1-55. Cronbach, L.J. (1972). Essentials of psychological testing. New York: Knight, D., Lattuca, L.R., Yin, A., Kremer, G., York, T., & Kyoung R.H. HarperCollins. (2012). An exploration of gender diversity in engineering programs: A CRUE: Conferencia de rectores de las Universidades españolas (2008). curriculum and instruction-based perspective. Journal of Women and La Universidad española en cifras [Spanish universities in figures]. Minorities in Science and Engineering, 18, 55-58. Observatorio Universitario. Little, T.D. (1997). Mean and Covariance structures (MACS) analyses Del Río, M. (2009). Logros de las mujeres en la arquitectura y la ingeniería of cross-cultural data: Practical and theoretical issues. Multivariate [Womens’ achievements in architecture and engineering]. Foro UPM. Behavioral Research, 32, 53-76. Díez, J.L. (2010). Proyecto Valentina [The Valentina project]. Presentation MacCallum, R.C., & Austin, J.T. (2000). Applications of structural equation at the 13th SIOU Meeting. Universidad Politécnica de Valencia. modeling in psychological research. Annual Review of Psychology, 51, Enginycat (2008). http://www.enginycat.cat. 201-226. Ferrando, F., Paleo, P., de la Flor, S., Urbina, C., & Gutiérrez-Colón, M. Medina, M.A. (2004). Mujeres... ¿Ingenieras?[Women... engineers?]. (2012). Estudio sobre la baja presencia de las mujeres en los estudios Round table at the congress 25 Years of Gender Studies. Universidad de ingeniería mecánica [Assessing the low percentage of women in Jaime I de Castellón. engineering degrees]. Presentation at the 19th National Congress of Millsap, R.E., & Meredith, W. (2007). Factorial invariance: Historical Mechanical Engineering. Universidad de Castellón. perspectives and new problems. In R. Cudeck & R.C. MacCallum Ferrando, P.J., & Demestre, J. (2008). Características de forma y contenido (Eds.), Factor analysis at 100 (pp. 131-152). Mahwah: LEA. que predicen la capacidad discriminativa en ítems de personalidad: un Muñiz, J. (1990). Teoría de respuesta a los ítems [Item response theory]. análisis basado en la Teoría de Respuesta a los Ítems [Characteristics Madrid: Pirámide. of content and form that predict discriminating power in personality Muthén, B., & Christoffersson, A. (1981). Simultaneous factor analysis of items: An IRT-based study]. Psicothema, 20, 851-856. dichotomous variables in several groups. Psychometrika, 46, 407-419. Hernández, A., & González-Romá, V. (2003). Evaluating the multiple- Muthén, L.K., & Muthén, B. (2007). Mplus user’s guide. Fifth edition. Los group mean and covariance structure analysis model for the detection of Angeles: Muthén & Muthén. 498
Distortions and gender-related differences in the perception of mechanical engineering in high school students National Survey of Recent College Graduates (NSRCG) (2013). http:// Spencer, S., Steele, C., & Quinn, D. (1999). Stereotype threats women’s www.nsf.gov/statistics/recentgrads/. Last visited: 25th April 2013. math performance. Journal of Experimental and Social Psychology, 35, Osgood, Ch. E., Suci, G.J., & Tannenbaum, P.H. (1957). The measurement 4-28. of meaning. University of Illinois Press. 499
You can also read